Computer Science Department
School of Computer Science, Carnegie Mellon University
A Practical System for Centralized Network Control
IP networking is a spectacular success, catalyzing the diffusion of data networking across academic institutions, governments, businesses, and homes worldwide. Yet, despite the fundamental importance of this infrastructure, today's networks are surprisingly fragile and increasingly difficult to configure, control, and maintain. As our dependence on data networking grows, so do the risks of security breaches, large-scale outages, and service disruptions.
We believe that the root of these problems lies in the complexity of the control and management planes – the software and protocols coordinating network elements – and particularly the way the decision logic and the distributed-systems issues are inexorably intertwined. The research community advocates a complete refactoring of the functionality and proposes a new architecture which they call "4D," after the architecture's four planes: decision, dissemination, discovery, and data. The 4D architecture pulls decision-making logic out of the network elements to create a logically centralized decision plane, where network-level objectives and policies are specified and enforced by direct configuration of state on individual network elements.
While the 4D vision is conceptually appealing, it has raised a wide range of practical concerns related to robustness, flexibility, scalability, and security. Our thesis is that "it is actually possible to build a 4D network that is as scalable and robust as traditional IP networks but greatly simplifies network control and management". To prove this thesis, we must address the following technical challenges:
1. What kind of decision-plane framework will enable the centralization and
composition of multiple network control functions for sophisticated network
We believe that answering the above questions is key to the successful deployment of 4D networks. In this dissertation, we tackle those challenges by building a 4D network control platform called Tesseract and demonstrating that Tesseract enables both simpler protocols that do not have to embed decision-making logic, and more powerful decision algorithms for implementing sophisticated goals. The main target of our work is to turn the revolutionary 4D concept into a practical working system.